Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[François Lacombe]

Hi

2018-02-27 13:10 GMT+01:00 Martin Koppenhoefer :

> I would facilitate things if navigable waterways are connected.
> Small streams should maybe only connect to the water body (lake).
>
> In theory you could do waterway routing similar to how pedestrian routing
> is done with squares in the cheapest version (around the square).
>

Given problem is there are sometimes waterways wich feed lakes from the
bottom, or under the shoreline
https://www.openstreetmap.org/way/562902636

Then I prefer to always connect even in lakes.


All the best

François
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Martin Koppenhoefer]

2018-02-23 11:35 GMT+01:00 Rory McCann :

> But then how far do you go? Should every stream be connected to the
> central river? e.g. what about here ( http://tools.geofabrik.de/osmi
> /?view=water=28.57869=-16.75136=11 )?
>
>

I would facilitate things if navigable waterways are connected.
Small streams should maybe only connect to the water body (lake).

In theory you could do waterway routing similar to how pedestrian routing
is done with squares in the cheapest version (around the square).

Cheers,
Martin
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Matej Lieskovský]

I'd like to reiterate that (much like riverbanks) the shoreline of a lake
can change far faster than the river channel. While the area maps the
"usual" extent (ignoring droughts and floods), the line is a very useful
abstraction.

Not connecting rivers into a network invalidates the idea of a network in
the first place and will lead to loss of data usefulness. This might lead
to wide rivers no longer getting a centerline (bacause what's the point if
everyone must work with the areas?) and will set a precedent that might get
used by (for example) highway:area proponents to do the same to the road
network. (No offence to the proponents of highway:area intended.) I'm
almost curious as to what mayhem would replacing roads with areas cause.

Should we not go the opposite direction and make sure that all rivers are
either tributaries, or flow into an ocean or a lake that would get tagged
with "no outflow"? I wouldn't be surprised to find out that someone already
does that.

On 23 February 2018 at 15:50, François Lacombe 
wrote:

> 2018-02-23 15:36 GMT+01:00 Rory McCann :
>
>> If OSM takes a "all rivers must be connected through lakes", then data
>> consumers have a simple job. If OSM says "some will and some won't", then
>> data consumers have to process the data to add intra-lake connections. If
>> they have to do it some of the time, why bother connecting *any* rivers?
>>
>
> IMHO rivers should always connect through lakes. I'm always mapping like
> this, no exception.
>
>
>> I think I'll change to not connecting rivers, unless it's very obvious,
>> and leaving data consumers to connect rivers themselves.
>>
> This may be a very hard task, especially if rivers don't share nodes witk
> lakes waterbody.
>
> François
>
> ___
> talk mailing list
> talk@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>
>
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[François Lacombe]

2018-02-23 15:36 GMT+01:00 Rory McCann :

> If OSM takes a "all rivers must be connected through lakes", then data
> consumers have a simple job. If OSM says "some will and some won't", then
> data consumers have to process the data to add intra-lake connections. If
> they have to do it some of the time, why bother connecting *any* rivers?
>

IMHO rivers should always connect through lakes. I'm always mapping like
this, no exception.


> I think I'll change to not connecting rivers, unless it's very obvious,
> and leaving data consumers to connect rivers themselves.
>
This may be a very hard task, especially if rivers don't share nodes witk
lakes waterbody.

François
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Rory McCann]

On 23/02/18 11:58, François Lacombe wrote:

If some rivers/streams shouldn't be connected, then some data consumers
will have to do an automatic connection anyway. When measuring water run
off and pollution, you probably want to know that "stuff going into
stream X will eventually get to point Y downstream" (right?).

I don't get this, which situation do you think of when you say " If some 
rivers/streams shouldn't be connected " ?


If OSM takes a "all rivers must be connected through lakes", then data 
consumers have a simple job. If OSM says "some will and some won't", 
then data consumers have to process the data to add intra-lake 
connections. If they have to do it some of the time, why bother 
connecting *any* rivers?


I think I'll change to not connecting rivers, unless it's very obvious, 
and leaving data consumers to connect rivers themselves.


___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Christoph Hormann]

On Friday 23 February 2018, Rory McCann wrote:
>
> But then how far do you go? Should every stream be connected to the
> central river? e.g. what about here (
> http://tools.geofabrik.de/osmi/?view=water=28.57869=-16.75136
>=11 )?
>
> If some rivers/streams shouldn't be connected, then some data
> consumers will have to do an automatic connection anyway. When
> measuring water run off and pollution, you probably want to know that
> "stuff going into stream X will eventually get to point Y downstream"
> (right?).

For simple formal connectivity that is essentially a routing problem.  
Not having connectivity on the waterway level will require you to use 
the water area data in analysis and this will more than double the data 
volume to process and if you need more than formal connectivity, i.e. 
quantitative information like path length or route geometry things 
become very expensive computationally.

> Connecting all means that large lakes will be full of a "skeleton" of
> joining rivers/streams, and a small 1km stream could get a lot
> longer.

You could detect the 'virtual' waterways as those within a non-river 
water area.  This is hampered by the problem that we mostly have no 
consistent distinction between river and lake areas in OSM (i.e. 
standing and flowing water areas).

-- 
Christoph Hormann
http://www.imagico.de/

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Maarten Deen]

On 2018-02-23 11:35, Rory McCann wrote:

On 23/02/18 06:53, Maarten Deen wrote:
I see nothing wrong with those examples, I would do it the same, 
especially if the rivers can be sailed on by boat. Then you absolutely 
need the rivers to be connected to a central river (or fairway) in the 
lake.


But then how far do you go? Should every stream be connected to the
central river? e.g. what about here (
http://tools.geofabrik.de/osmi/?view=water=28.57869=-16.75136=11
)?


Mappers discretion. I'm not advocating a "must" in this case. I'm saying 
it's not a bad thing if it is done and in some circumstances it is 
preferrable.
See it as roads connecting to a square. As long as routing over areas is 
not commonplace it is acceptable and even preferrable to lay roads over 
the square.



If some rivers/streams shouldn't be connected, then some data consumers
will have to do an automatic connection anyway. When measuring water 
run

off and pollution, you probably want to know that "stuff going into
stream X will eventually get to point Y downstream" (right?).


Yes. OSM gets used (maybe misused) for all kinds of purposes, but a 
hydrological analysis would be impossible if streams that are physically 
connected are not so in OSM.



Connecting all means that large lakes will be full of a "skeleton" of
joining rivers/streams, and a small 1km stream could get a lot longer.


Good point, but can be solved by using relations for the streams and not 
including the part of the stream that runs through the lake. In any 
case, I would think it strange to still use the the name of the stream 
on the part that runs through the lake.


Maarten

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[François Lacombe]

 (Sorry Rory, resent this to Talk ML)

2018-02-23 11:35 GMT+01:00 Rory McCann :

> On 23/02/18 06:53, Maarten Deen wrote:
>
>> I see nothing wrong with those examples, I would do it the same,
>> especially if the rivers can be sailed on by boat. Then you absolutely need
>> the rivers to be connected to a central river (or fairway) in the lake.
>>
>
> But then how far do you go? Should every stream be connected to the
> central river? e.g. what about here ( http://tools.geofabrik.de/osmi
> /?view=water=28.57869=-16.75136=11 )?
>

Yes: http://tools.geofabrik.de/osmi/?view=water=28.57869;
lat=-16.75136=11


> If some rivers/streams shouldn't be connected, then some data consumers
> will have to do an automatic connection anyway. When measuring water run
> off and pollution, you probably want to know that "stuff going into
> stream X will eventually get to point Y downstream" (right?).
>

I don't get this, which situation do you think of when you say " If some
rivers/streams shouldn't be connected " ?


> Connecting all means that large lakes will be full of a "skeleton" of
> joining rivers/streams, and a small 1km stream could get a lot longer.

Yes they do http://tools.geofabrik.de/osmi/?view=water=28.57869;
lat=-16.75136=11

This can be solved by removing waterways sections which intersect with
lakes water body.
It could be done on consumer purpose for a particular usage.
Topology is also a really useful data and waterways should definetly
connect downstream sections.

2018-02-23 11:45 GMT+01:00 Joseph Reeves :

> Slightly off topic, but I was recently wondering if there was a waterway
> routing tool available? As in, I'd like to click a point in a waterway and
> have the downstream route plotted, presumably to the sea. It appears to me
> that a tool like that could be useful in this discussion?
>

You can achieve this with OSRM with a proper routing profile, or even
pg-routing if you are at ease with it.

All the best

François
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Joseph Reeves]

Hi all,

Slightly off topic, but I was recently wondering if there was a waterway
routing tool available? As in, I'd like to click a point in a waterway and
have the downstream route plotted, presumably to the sea. It appears to me
that a tool like that could be useful in this discussion?

Despite my best efforts, I keep finding river drainage basins fascinating :)

Cheers, Joseph



On 23 February 2018 at 10:35, Rory McCann  wrote:

> On 23/02/18 06:53, Maarten Deen wrote:
>
>> I see nothing wrong with those examples, I would do it the same,
>> especially if the rivers can be sailed on by boat. Then you absolutely need
>> the rivers to be connected to a central river (or fairway) in the lake.
>>
>
> But then how far do you go? Should every stream be connected to the
> central river? e.g. what about here ( http://tools.geofabrik.de/osmi
> /?view=water=28.57869=-16.75136=11 )?
>
> If some rivers/streams shouldn't be connected, then some data consumers
> will have to do an automatic connection anyway. When measuring water run
> off and pollution, you probably want to know that "stuff going into
> stream X will eventually get to point Y downstream" (right?).
>
> Connecting all means that large lakes will be full of a "skeleton" of
> joining rivers/streams, and a small 1km stream could get a lot longer.
>
>
>
> ___
> talk mailing list
> talk@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Rory McCann]

On 23/02/18 06:53, Maarten Deen wrote:
I see nothing wrong with those examples, I would do it the same, 
especially if the rivers can be sailed on by boat. Then you absolutely 
need the rivers to be connected to a central river (or fairway) in the 
lake.


But then how far do you go? Should every stream be connected to the
central river? e.g. what about here ( 
http://tools.geofabrik.de/osmi/?view=water=28.57869=-16.75136=11 
)?


If some rivers/streams shouldn't be connected, then some data consumers
will have to do an automatic connection anyway. When measuring water run
off and pollution, you probably want to know that "stuff going into
stream X will eventually get to point Y downstream" (right?).

Connecting all means that large lakes will be full of a "skeleton" of
joining rivers/streams, and a small 1km stream could get a lot longer.


___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[François Lacombe]

2018-02-23 6:53 GMT+01:00 Maarten Deen :

> On 2018-02-22 22:59, Rory McCann wrote:
>
> If you asked someone "Where does this river end?" they'd probably point
>> to where it joins the lake. Connecting the river to the "central river"
>> breaks this. And it can result in odd long ways. I might have gone a
>> little OTT here (
>> http://tools.geofabrik.de/osmi/?view=water=27.98062=
>> -16.95179=11
>> ) or here (
>> http://tools.geofabrik.de/osmi/?view=water=-8.26705=
>> 52.99047=11
>> ).
>>
>
> I see nothing wrong with those examples, I would do it the same,
> especially if the rivers can be sailed on by boat. Then you absolutely need
> the rivers to be connected to a central river (or fairway) in the lake.
>

I agree too and this is how I use to map rivers in lakes.

This is great thing to have comprehensive topology for waterways.

François
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Matej Lieskovský]

Greetings!

I saw a nice argument somewhere about the lines for larger rivers that
might be applicable here:
Imagine if the water level dropped significantly - the areas would be off
(but should be kept at their normal level unless the change is likely to be
reasonably permanent) but the lines should stay reasonably accurate,
mapping the main channel where the water will keep flowing. Of course,
unless the water drops it might not be possible to find out where such a
channel is exactly, so the map usually contains our best guesses.

Happy mapping!

On 23 February 2018 at 06:53, Maarten Deen  wrote:

> On 2018-02-22 22:59, Rory McCann wrote:
>
>> Hi mappers,
>>
>> What's the best way to map rivers that flow into lakes, especially when
>> another river flows through it? Should they be connected?
>>
>> When a river flows through a lake, you can map a waterway=river way
>> through it, to be "topoligcally complete". Or would it be better to add
>> ways (w/o waterway tag) to the river relation?
>>
>> When a tributary river joins another, join the central waterway=river
>> ways together. But what if a river drains into a lake with a  "central
>> river" through it? Should you connect that river to the central river?
>> It makes topological sense.
>>
>> If you asked someone "Where does this river end?" they'd probably point
>> to where it joins the lake. Connecting the river to the "central river"
>> breaks this. And it can result in odd long ways. I might have gone a
>> little OTT here (
>> http://tools.geofabrik.de/osmi/?view=water=27.98062=
>> -16.95179=11
>> ) or here (
>> http://tools.geofabrik.de/osmi/?view=water=-8.26705=
>> 52.99047=11
>> ).
>>
>
> I see nothing wrong with those examples, I would do it the same,
> especially if the rivers can be sailed on by boat. Then you absolutely need
> the rivers to be connected to a central river (or fairway) in the lake.
>
> Maarten
>
>
> ___
> talk mailing list
> talk@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk
>
___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Maarten Deen]

On 2018-02-22 22:59, Rory McCann wrote:

Hi mappers,

What's the best way to map rivers that flow into lakes, especially when
another river flows through it? Should they be connected?

When a river flows through a lake, you can map a waterway=river way
through it, to be "topoligcally complete". Or would it be better to add
ways (w/o waterway tag) to the river relation?

When a tributary river joins another, join the central waterway=river
ways together. But what if a river drains into a lake with a  "central
river" through it? Should you connect that river to the central river?
It makes topological sense.

If you asked someone "Where does this river end?" they'd probably point
to where it joins the lake. Connecting the river to the "central river"
breaks this. And it can result in odd long ways. I might have gone a
little OTT here (
http://tools.geofabrik.de/osmi/?view=water=27.98062=-16.95179=11
) or here (
http://tools.geofabrik.de/osmi/?view=water=-8.26705=52.99047=11
).


I see nothing wrong with those examples, I would do it the same, 
especially if the rivers can be sailed on by boat. Then you absolutely 
need the rivers to be connected to a central river (or fairway) in the 
lake.


Maarten

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

Re: [OSM-talk] Mapping rivers that flow into/through lakes?

[Christoph Hormann]

On Thursday 22 February 2018, Rory McCann wrote:
>
> What's the best way to map rivers that flow into lakes, especially
> when another river flows through it? Should they be connected?
>
> [...]

This has been discussed in the past occasionally and IIRC there was 
never full agreement on the matter.  A few things that are important to 
consider:

* The question if a river flows through a lake or if it enters a lake 
and ends there is not generally something that can be answered 
verifiably.
* In well mapped areas where a lake has one large outflow and only one 
or a few similarily large inflows they are often connected.
* In most cases with larger lakes smaller inflows are not connected.
* In contrast to riverbank polygons where there is in principle a well 
established rule where to place the waterway (the thalweg) there is no 
established rule how rivers that are connected within a lake are to be 
placed or connected.
* There is also no established rule if waterways within a lake get the 
name and other tags from the tributaries they connect or if they are to 
be without name tag.  Both variants are common.
* The vast majority of waterbody imports import without connectivity, 
even if connected data is available like in the US and Canada.

Part of the problem is of course that the task of generating waterway 
connectivity within a water area or supplementing an incomplete 
connectivity are cumbersome mechanical tasks.  So the ultimate question 
probably is if this is something editors should provide support to 
produce automatically or if it is something that data users should 
generate automatically as needed.

-- 
Christoph Hormann
http://www.imagico.de/

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk

[OSM-talk] Mapping rivers that flow into/through lakes?

[Rory McCann]

Hi mappers,

What's the best way to map rivers that flow into lakes, especially when
another river flows through it? Should they be connected?

When a river flows through a lake, you can map a waterway=river way
through it, to be "topoligcally complete". Or would it be better to add
ways (w/o waterway tag) to the river relation?

When a tributary river joins another, join the central waterway=river
ways together. But what if a river drains into a lake with a  "central
river" through it? Should you connect that river to the central river?
It makes topological sense.

If you asked someone "Where does this river end?" they'd probably point
to where it joins the lake. Connecting the river to the "central river"
breaks this. And it can result in odd long ways. I might have gone a
little OTT here (
http://tools.geofabrik.de/osmi/?view=water=27.98062=-16.95179=11
) or here (
http://tools.geofabrik.de/osmi/?view=water=-8.26705=52.99047=11
).

Without joining the ways, then a data consumer will have to do
complicated extra processing to deduce that one river "flows into" the
other? (River X ends on the shore of a lake, River Y flows through the
lake, so connect X to Y). Is that "good enough"?

Thoughts?

Rory

___
talk mailing list
talk@openstreetmap.org
https://lists.openstreetmap.org/listinfo/talk